IDSgrep (overview, manual, academic paper) is a tool written by Matthew Skala for running "grep"-like text searches in databases of Chinese characters.
This repository contains an beta-state port to Javascript via Emscripten.
This means that a useful C program now runs in your browser's Javascript engine, having gone through a cross-compilation tool known as Emscripten. This may be the closest one can get to magic, for the next few minutes at least.
Geek's note: more specifically, the C code was compiled to LLVM intermediate representation (IR) by Clang, the C compiler, which Emscripten then compiled to Javascript.
Browse it now at http://fasiha.github.io/idsgrep-emscripten/! Run your Han character queries against the CVKVI and KanjiVG databases!
For information on what kind of queries can be run, see the IDSgrep manual, as well as the sample queries that are run when you first load the page. More tutorial material will soon be added.
After getting the IDSgrep source with ancillary databases and setting up Emscripten, I run the configure script using Emscripten's emconfigure wrapper:
$ emconfigure ./configure --disable-docs --disable-colour-build --disable-silent-rules
Then, because emconfigure doesn't seem to respect CFLAGS command-line arguments, I have to manually edit the resulting Makefile (included in this repository, but will be overwritten by the above) to include
CFLAGS = -O0 --llvm-opts 0Next, execute the Makefile within Emscripten:
$ emmake make
Rename the LLVM intermediate representation (IR) bitcode to have .bc extension so emcc knows what to do with it, and then build it for the web!:
$ cp idsgrep idsgrep.bc
$ emcc -O0 idsgrep.bc -o index.html --preload-file dict/cjkvi-j.eids --preload-file dict/cjkvi-j.bvec --preload-file dict/kanjivg.bvec --preload-file dict/kanjivg.eids -s EXPORTED_FUNCTIONS="['_core_process', '_core_init', '_core_test', '_core_cook', '_set_output_recipe']" -s MEMFS_APPEND_TO_TYPED_ARRAYS=1
The output is hosted at http://fasiha.github.io/idsgrep-emscripten/.
- It's unclear why mnemonics like "[lr]" and "(anything)" mentioned in the IDSgrep manual, pages 19--20. I think this has something to do with the cooking process. This is a work-in-progress.